Automated System Recovery Tool

[James Antill]

On Wed, 2009-04-01 at 11:52 -0700, J C wrote:
> 
> Title:  Automated System Recovery Tool  
> Student:  Jeff Chandler  
> Abstract:  
> An automated system
> recovery tool is needed for average users to feel secure using Fedora
> as their primary Desktop OS. Fedora and other Gnu/Linux distros can be
> rendered useless to the non-technical user by a failed binary driver
> install (Video cards, wireless ethernet, etc.) or malicious programs.
> For these users, there is no easy fix available.

 How are you going to run your daemon (let alone the UI to command it)
if arbitrary parts of the system are broken?

>  I propose a daemon
> that tracks changes in the filesystem and can restore installed
> programs and configuration to a previous point in time when the system
> was working. Suspect programs would be cordoned off into a secure
> area.

 So do you only protect packages, and not all files on disk?
 If more than packages how do you tell which ones (reverting a config.
file might require reverting a package, and it's much more likely the
other way around).
 If only package data, then you should really be monitoring/using the
"yum history" database (not written, yet, but planned for this year).

[...]
>    A daemon that would track changes in the filesystem, installed
> packages and versions, and configuration information.  In the event of
> a system failure or corruption (by a binary driver, user error,...)the
> Automated System Recovery Tool could be used to restore the system to
> a
> previous time.  The daemon would also keep a copy of non-standard (not
> available from the standard repos) packages for reinstall. The tool
> could also be used as a System Backup utility to store all system and
> user data in a compressed file in case of a total system or hardware
> failure.

 A backup utility like this would be a huge undertaking, on it's own, I
think (even if you ignored the packaging aspect).

[...]
> * A rough timeline for your progress
>   1.) Implement daemon to record system changes to a recovery file. 1
> week
>   2.) Implement Graphical Interface for user interaction. 1 week
>   3.) Implement System Recovery backend 2 weeks
>   4.) Test and Debug. 2 weeks
> * Any other details you feel we should consider

 IMO the above is "optimistic".

-- 
James Antill <james at fedoraproject.org>
Fedora